Retrieval-Augmented Generation pipeline over internal documents using ada-002 embeddings, PostgreSQL + pgvector, and LangChain orchestration. Reduced search time by 40%.
Sergio Daniel Casas — software engineer with 6+ years of experience building highly available web applications, infrastructure automation, and AI-powered solutions.
I focus on the intersection of LLMs, RAG, and resilient production architectures. I value well-designed systems, observability, and honest documentation. Code should speak for itself.
# python · go · aws · gcp
# llm · rag · agents · embeddings
# docker · k8s · terraform · linux
Retrieval-Augmented Generation pipeline over internal documents using ada-002 embeddings, PostgreSQL + pgvector, and LangChain orchestration. Reduced search time by 40%.
Gateway with JWT authentication, rate limiting, Prometheus observability, and dynamic routing. Deployed on Kubernetes (GKE) with zero-downtime rollouts.
SaaS template with Django + Celery + Redis, multi-tenant authentication, Stripe billing, and automated deployment from GitHub Actions to AWS ECS.
Terraform modules for reproducible environments on AWS and GCP: VPC, EKS/GKE, RDS, S3, and monitoring stack. Reduced new environment setup time by 70%.
Asynchronous processing system with FastAPI, Celery, and Redis. Real-time monitoring, automatic retries, and dead-letter queues. Up to 10k tasks/min.